Post on 25-Apr-2020
Efficiency Wages, Unemployment
and Macroeconomic Policy
Jim Malley
University of Glasgow
Hassan Molana University of Dundee
20 March 2002
ABSTRACT: We provide empirical evidence from a number of European countries, which shows that unemployment and output are positively related when unemployment is low and inversely related when unemployment is high. We then construct a stylised macro-model with goods and labour market imperfections to show that the economy can rationally operate at an inefficient equilibrium in the neighbourhood of which the relationship between output and unemployment is positive. Our results suggest that circumstances exist in which market imperfections pose serious obstacles to the smooth working of expansionary and/or stabilization policies whose final aim is to improve welfare. KEYWORDS: Efficiency wages; effort supply; monopolistic competition; multiple equilibria; stability; fiscal multiplier JEL CLASSIFICATION: E62, J41, H3 Correspondence: Hassan Molana, Department of Economics, University of Dundee, Dundee DD1 4HN, UK. Tel: (00)44-(0)1382-344375; Fax: (00)44-(0)1382-344691; E-mail: h.h.molana@dundee.ac.uk. We would like to thank Jonas Agell, Torben Andersen, Alessandro Cigno, Laszlo Goerke, Bertil Holmlund, Lawrence Kahn, Thomas Moutos and participants at the CESifo Area Conference on Employment and Social Protection (May 2001, Munich) and at the Business Cycle Theory session of the European Economic Association Congress (August 2001, Lausanne) for helpful comments on earlier versions of this paper.
1
1. Introduction
In the last few decades, the industrialised nations of Europe have been subjected to a
variety of external and policy-induced demand shocks while simultaneously experiencing
significant changes in their labour productivity and employment. Meanwhile, some leading
economists have argued that persistent involuntary unemployment may give rise to
externalities that could be exploited by economic agents. A much-discussed example in this
context is for price-setting firms to use high or rising unemployment as a device to deter
shirking. Lindbeck (1992) suggests that in this setting, macroeconomic policy interventions
may produce unexpected consequences: "In the context of a nonmarket-clearing labour
market, it is certainly reasonable to regard unemployment, in particular highly persistent
unemployment, as a major macroeconomic distortion. There is therefore a potential case for
policy actions, provided such actions do not create more problems than they solve.
Experience in many countries suggests that the latter reservation is not trivial."
To more fully explore the extent to which these concerns are warranted, we first
examine the aggregate data on unemployment and output for a number of European
countries. Our empirical evidence shows that it is not uncommon for output and
unemployment to be related according to a humped-shape relationship. More specifically,
we find that changes in output and unemployment are correlated negatively (positively) when
the latter is relatively high (low). It is when these variables co-vary positively that an
expansionary shock may yield counterintuitive effects. We next develop a theoretical model
that shows such a relationship can result when labour and goods market operate under certain
plausible conditions. An important policy insight that emerges from this paper is that an
exogenous stimulation of aggregate demand can only raise output and reduce unemployment
provided the economy is operating relatively efficiently. The intuition for this lies in the
supply side nonlinearities that could give rise to multiple equilibria. We show that when an
2
economy is trapped in an inefficient equilibrium, positive demand shocks can lead,
perversely, to an increase in unemployment.
Other recent examples of related theoretical research, which examines the link
between European unemployment and productivity, include Malley and Moutos (2001), Leith
and Li (2001), Daveri and Tabellini (2000), Blanchard (1998), Caballero and Hammour
(1998a,b), Gordon (1997) and Manning (1992). However, none of these studies explores the
link between unemployment and output arising from both labour and product market
imperfections.
On the empirical side, a large number of studies have examined the behaviour of
labour productivity in the industrialised countries and provide indisputable evidence
regarding the way in which labour productivity has changed over the last few decades.
Recent examples include Disney, et al. (2000), Barnes and Haskel (2000), Marini and
Scaramozzino (2000), Bart van Ark et al. (2000) and Sala-i-Martin (1996). The evidence
provided in these studies is usually interpreted using one of the micro-theory based
explanations underlying the behaviour of labour productivity. These may, in general, be
divided into two categories. The first concentrates on the productivity gains that can be
realised through: i) improved skill due to training; ii) increased efficiency due to progress in
management and restructuring; and iii) rising physical productivity of other factors of
production due to R&D, etc.. The second category emphasises market forces and sees
competition and market selection as the main motivation behind the rise in efficiency. The
separating line between these two accounts is not very clear in the sense that the second will
have to be achieved through the first when the economy is operating efficiently. However, if
the economy happens to be in an inefficient phase, market forces can act directly without
3
having to induce any of the factors in the first category. The efficiency wage hypothesis is a
typical example of this case and will be used in this paper to illustrate the point.
The rest of the paper proceeds as follows. Section 2 examines data from a number of
European countries on output and unemployment and reports the evidence on how
movements in output are matched with changes in unemployment. Section 3 outlines a
theoretical model based on the efficiency wage hypothesis and shows that the production side
of the economy – consisting of monopolistically competitive firms that set prices and offer
efficiency wages to maximise profits – can give rise to a non-linear equilibrium relationship
between output and unemployment, consistent with the evidence reported in Section 2.
Section 3 concludes the paper and finally the Appendix outlines the derivation of the effort
function we utilise in our theoretical model.
2. Output and Unemployment: Some Stylised Facts
Our aim in this section is to explore the way changes in unemployment and output are
related to each other empirically. However, it would be helpful if, a priori, we postulate an
equilibrium that sustains involuntary unemployment and allow for the latter to affect
workers’ effort supply. In this case, because there is a causal relationship between the level
of unemployment and the productivity of the employed, a total change in output can be
decomposed into changes due to employment and productivity. As a result, a sufficient
condition for an expansionary demand shock to raise both output and employment is that the
resulting fall in unemployment does not induce a fall in productivity of the employed to such
an extent that it eliminates the effect of the rise in employment. Defining aggregate labour
productivity as q=Y/L where Y, L and q respectively denote output, employment and
productivity, and noting that dq=(dY-qdL)/L, it is clear that any of the six cases outlined in
4
Table 1 could, in principle, occur in the aggregate (see Barnes and Haskel, 2000, for evidence
at plant level).
Table 1. Simultaneous Changes in Labour Productivity,
Output and Employment. Change in
Labour Productivity
Change in Employment
Change in Output
Change in Unemployment
Rate dL<0 dY>0 du>0 dL>0 dY>0 du<0
Rising Productivity
dq>0 dL<0 dY<0 du>0 dL>0 dY<0 du<0 dL<0 dY<0 du>0
Falling Productivity
dq<0 dL>0 dY>0 du<0 The last column of Table 1 shows the implied changes in the unemployment rate
(based on the approximation that the labour force is constant). This discussion clearly
suggests that it is a distinct possibility that output and unemployment can fall or rise
simultaneously. While the cases in which changes in output and unemployment have
opposite signs can be easily explained by a variety of standard theories, a convincing
macroeconomic theory capable of predicting why these variables fall or rise simultaneously is
more elusive. To obtain a more realistic indication of whether output and unemployment
simultaneously move in the same direction, in Table 2 we examine quarterly data for a cross
section of 10 European countries, chosen to reflect a wide range of industrial structures as
well as macroeconomic and labour market experiences over the last few decades.
Table 2 shows, for each country, the directions of quarterly changes as a proportion of
the entire sample for which changes in output and unemployment have the same sign, that is
]0du&0dY[ tst >>± and ]0du&0dY[ tst <<± for s=0,2,4. The results corresponding to
contemporaneous changes (middle column) indicate that for a substantial and statistically
significant proportion of the sample (i.e. at least 35% of the sample for all countries) the sign
combinations show output and unemployment moving in the same direction. It is also clear
5
Table 2. Directions of Quarterly Changes in Output, Y, & Unemployment Rate, u;
(occurrences as percentage of the sample size, n)
C
ountry
Sam
ple
dYt-4 >0 &
dut >0
and dY
t-4 <0 & du
t <0
dYt-2 >0 &
dut >0
and dY
t-2 <0 & du
t <0
dYt >0 &
dut >0
and dY
t <0 & du
t <0
dYt+2 >0 &
dut >0
and dY
t+2 <0 & du
t <0
dYt+4 >0 &
dut >0
and dY
t+4 <0 & du
t <0
BEL 60:2-97:4 (n=151)
0.456 0.423 0.417 0.423 0.463
DEU 60:2-89:4 (n=119)
0.461 0.385 0.345 0.402 0.461
ESP 61:1-98:4 (n=152)
0.493 0.500 0.480 0.473 0.493
FRA 65:1-97:4 (n=132)
0.609 0.592 0.553 0.638 0.656
GBR 60:2-98:3 (n=154)
0.433 0.414 0.370 0.461 0.547
IRE 60:2-97:4 (n=151)
0.497 0.463 0.430 0.450 0.456
ITA 60:2-98:3 (n=154)
0.487 0.559 0.513 0.546 0.560
NLD 69:2-97:4 (n=115)
0.414 0.416 0.400 0.416 0.414
PRT 60:2-97:4 (n=151)
0.578 0.597 0.629 0.671 0.694
SWE 60:2-98:3 (n=154)
0.473 0.342 0.357 0.467 0.460
The number of observations, n, and the sample period correspond to the contemporaneous changes in the natural logarithm of real GDP (market prices), Y, and the unemployment rate, u. Based on a one-sample 2-tailed t-test the mean number occurrences of (dYt±s>0 & dut>0) and (dYt±s<0 & dut<0) are significantly different from zero at the 0.01 level of significance. This result also holds across all leads, lags and countries. Countries are defined as follows: Belgium (BEL), West Germany (DEU), Spain (ESP), France (FRA), UK (GBR), Ireland (IRE), Italy (ITA), Netherlands (NLD), Portugal (PRT), Sweden (SWE); Y and u are obtained from the OECD Business Sector Database.
from Table 2 that these findings hold for lagged and led changes, which we have considered
in order to capture the variations over business cycle. Moreover, when the non-
contemporaneous changes are considered, the proportion of periods where Y and u are
positively related increases for virtually all other cases considered1. Given the results in
Table 2, it seems fair to conclude that theories disregarding this possibility – by employing
models with market structures which are only capable of generating the prediction that Y and
6
u are negatively related – can only be of limited use when analysing the potential effects of
macroeconomic policies.
The evidence in Table 2 pertaining to changes in output and unemployment having
the same sign is potentially consistent with a number of alternative explanations, e.g. i) an
exogenous increase the labour force participation; ii) the net result of simultaneous
exogenous shocks to both aggregate supply and demand which would move output and
unemployment simultaneously in the same direction; or iii) the theories of creative
destruction (see, for examples, Aghion and Howitt, 1992; 1994). However, if the periods in
which output and unemployment are positively related occur on a systematic basis, the above
explanations will prove inadequate since they all rely on random impulse mechanisms. We
therefore next focus our analysis on this aspect of the evidence, i.e. a regular occurrence of
positive and negative correlations between changes in output and unemployment. To do so,
we first test to see if the occurrences of ]0du&0dY[ tt >> and ]0du&0dY[ tt <<
reported in Table 2 are random. To this end we employ a simple ‘runs test’ which is a one-
sample nonparametric test for randomness in a dichotomous variable. Given that a run is
defined as any sequence of cases having the same value, we assign 1 to those periods where
[dYt>0 & dut>0] and [dYt<0 & dut<0] and 0 to all other periods. The total number of runs in
the sample is a measure of randomness in the order of the cases; too many or too few runs
can suggest a non-random, or dependent, ordering. The results are reported in Table 3 below
where the rows in bold indicate the countries (i.e. Italy and Sweden) for which we are unable
to reject the null hypothesis of randomness in the runs. Accordingly, these countries are
excluded from further analysis below where we will focus on exploring the existence and
1 In the empirical analysis, which follows, we shall concentrate on the contemporaneous changes only since
this case clearly does not over-record the proportion of periods when Y and u are moving in the same direction.
7
nature of the systematic relationship between changes in unemployment and output, as
supported by the majority of cases.
Table 3. Runs Tests for the Significant Systematic Occurrences of
[dYt>0 & dut>0] and [dYt<0 & dut<0] Cases. Country T.V. Cases < T.V. Cases ≥≥≥≥T.V. Total
Cases No. of Runs
T.S.V. A .S.
BEL 0.417 88 63 151 25 -8.301 0.000 DEU 0.345 78 41 119 43 -2.397 0.017 ESP 0.480 79 73 152 47 -4.871 0.000 FRA 0.553 59 73 132 50 -2.874 0.004 GBR 0.370 97 57 154 58 -2.568 0.010 IRE 0.430 86 65 151 29 -7.668 0.000 ITA 0.513 75 79 154 76 -0.315 0.753 NLD 0.400 69 46 115 35 -4.138 0.000 PRT 0.629 56 95 151 42 -5.158 0.000 SWE 0.357 99 55 154 62 -1.711 0.087
T.V. is the Test Value (mean cut point); T.S.V. is the value of the test statistic; A.S. is the 2-tailed asymptotic significance level.
At this stage it is helpful to compare summary measures of central tendency of the
unemployment rate across the two different states, i.e. state 1: ]0du&0dY[ tt >> and
]0du&0dY[ tt << and state 0: ]0du&0dY[ tt <> and [ 0 & 0]t tdY du< > . These
simple comparisons are reported in Table 4 below and show clearly that states 0 and 1 are
likely to occur at high and low levels of unemployment, respectively. With the exception of
German data, the evidence shows a clear tendency for unemployment to be lower when
output and unemployment are positively related and higher when they are negatively related.
The clear pattern that emerges from the analysis, summarised as
State 0: ]0du&0dY[ tt <> and [ 0 & 0]t tdY du< > are more likely to occur when ut is relatively high
State 1: ]0du&0dY[ tt >> and ]0du&0dY[ tt << are more likely to occur
when ut is relatively low; and
8
Table 4. Comparing Mean and Median Unemployment Rates for the
Periods when Y and u are positively and negatively related
Country
Mean for State 1 Periods
Mean for State 0 Periods
Difference Between Means
Median for State 1 Periods
Median for State 0 Periods
Difference Between Medians
BEL 6.655 7.345 -0.689 7.066 8.847 -1.781 DEU 3.503 3.267 0.236 3.246 2.909 0.337 ESP 10.436 11.684 -1.248 7.603 15.351 -7.748 FRA 6.285 7.609 -1.323 5.661 9.002 -3.341 GBR 5.615 5.803 -0.187 4.419 5.914 -1.495 IRE 9.350 9.611 -0.261 7.765 8.310 -0.545 NLD 4.911 5.965 -1.054 4.071 5.856 -1.785 PRT 5.962 6.789 -0.828 5.593 7.106 -1.513
can be used to postulate a behavioural, or theoretical, structure that could embrace the above
findings. For instance, as we shall see later, this type of systematic pattern in data is
consistent with the theoretical implications of the efficiency wage hypothesis, which predicts
a causal relationship between output and unemployment through the effect of the latter on
productivity of workers. Of the 10 European countries examined above only Italy, Sweden
and Germany do not match this prediction, even though as seen from Table 2 they too have a
significant number of occurrences of output and unemployment moving in the same
direction. This discrepancy may be due to the differences in the way labour markets function
in these countries, e.g. use of guest labour, style of unionisation etc.. However, further
investigation of these results is beyond the scope of this paper, as we do not attempt to
empirically establish the general validity of any particular theory. Our primary purpose is to
understand the regularities in data and to develop a theory capable of explaining them.
As a final attempt to further describe the stylised relationship between output and
unemployment, suggested by the above data analysis, we estimate by GMM a quadratic
relationship between de-trended output and unemployment. The results of this estimation for
the remaining seven countries in Table 4 are given in Table 5 below. They show clear
9
support for a humped-shaped relationship between output and unemployment. As a tentative
measure, we have also calculated the level of unemployment, denoted by u , at which output
reaches its maximum value. Note that, according to the behaviour postulated above, u is the
threshold level of unemployment, which separates State 1 from State 0.
Table 5. GMM Estimates of ( )( )2
0 1 2 0t t t t tE Y u uβ β β − + + = z%
C
ountry
1�β
2�β
2
1�2
�u
ββ−=
J-Statistic
BEL
0.0981 (0.0258)
-0.0060 (0.0017)
8.12
5.98 [0.426]
ESP
0.0681 (0.0310)
-0.0026 (0.0013)
13.31
8.67 [0.193]
FRA
0.1221 (0.0469)
-0.0074 (0.0035)
8.27
2.98 [0.812]
GBR
0.1483 (0.0271)
-0.0106 (0.0021)
7.01
6.35 [0.385]
IRE
0.1676 (0.0747)
-0.0068 (0.0034)
12.31
7.73 [0.258]
NLD
0.1094 (0.0296)
-0.0093 (0.0027)
5.87
7.53 [0.274]
PRT
0.3121 (0.1302)
-0.0229 (0.0099)
6.83
9.16 [0.165]
Y~ is de-trended ln(Y). Four lags of Y~ and u were used as instruments in vector z. Estimation method was based on Quadratic Barlett Kernel using 8 autocorrelation terms. Numbers in parentheses are the estimated HAC standard errors. Numbers in square brackets are the corresponding p-values. The J-Statistic is for Hansen’s test of the validity of the overidentifying restrictions.
3. Output and Unemployment: Theory
In this section we examine whether the relationship deduced from data in the previous
section can be derived from a stylised theoretical macro-model – e.g. of the type outlined by
Blanchard and Kiyotaki (1987). We therefore construct a suitable version of the latter, which
allows for unemployment to be sustained in equilibrium. To incorporate this modification we
endow the model with market imperfections in both the labour and the goods market by
10
allowing for firms to be monopolistically competitive and to reward workers’ effort by
paying efficiency wages.
Following the work of Shapiro and Stiglitz (1984) and Yellen (1984), a number of
models have employed some version of the efficiency wage hypothesis to study various
aspects of macroeconomic activity. Examples can found in: Agénor and Aizenman (1999)
and Rebitzer and Taylor (1995) on fiscal and labour market policies; Andersen and
Rasmussen (1999), Pisauro (1991) and Carter (1999) on the role of the tax system; Leamer
(1999) on specialisation; Albrecht and Vroman (1996), Fehr (1991) and Smidt-Sørensen
(1990) on properties of labour demand; and Smidt-Sørensen (1991) on working hours. In this
paper we employ a standard version of the hypothesis which postulates an effort supply
function that is derived by workers maximising the expected utility from work as explained in
the Appendix.
The model is static and describes an economy with three types of agents: firms,
households and a government. Firms are monopolistically competitive and each produce a
variety of a horizontally differentiated product using labour as input with an increasing
returns to scale technology. Households are endowed with a unit of labour, which they
supply inelastically. Unemployed households receive a benefit transfer from the government.
The government revenue, raised by taxing the households, is used to subsidise the
unemployed and to pay for government consumption. The final good in the economy is the
Dixit-Stiglitz CES bundle of horizontally differentiated varieties produced by identical firms
described later.
The demand side of the model consists of the households’ and government’s
consumption. The latter is given by
11
PGdjgPNj
jj =∫∈
, (1)
where j is the index denoting a variety of the differentiated good, N is the mass of varieties
on offer and Pj and gj are the price and quantity of variety j. It is straightforward to show that
=
−
NG
PP
g jj
ε
, (2)
where G is the corresponding CES bundle with the constant elasticity of substitution2
between any two varieties s>1,
)]s/1(1/[1
Nj
)s/1(1j
s/1 djgNG−
∈
−−
= ∫ ,
and P is the price index dual to G; (2) maximises G above subject to (1). The government
expenditure comprises (1) and the unemployment benefit payments B per unemployed
worker/household. This expenditure is financed by a lump sum tax3, T, which, together with
the normalisation of the number of households to unity – on the assumption that each
household is endowed with one unit of labour – gives rise to the following budget constraint
PG + uB = T. (3)
Each household is endowed with initial money holdings M and receives distributed
profits Π. In addition, it also supplies, inelastically, its unit of labour and at any point in time
it can either be employed or be unemployed. When employed, a typical household works for
a firm j, supplying the effort level ej>0 and earning nominal wage Wj. If unemployed, it
receives from the government the nominal unemployment benefit B at no effort. Dropping
the distinction between firms and setting profit income to zero (anticipating the symmetric
2 See Molana and Zhang (2001) for a study of the role of a variable elasticity of substitution in the context of
fiscal policy effectiveness. Note also that, following the common practice the CES bundle is normalised by the mass of varieties, N, to switch off the variety effect in the aggregate.
12
equilibrium and elimination of profits through a free entry and exit process), a household’s
budget constraint is
−+−+
=+unemployed,TMB
employed,TMWMPC (4)
where M is the desired stock of money, C is the aggregate CES bundle of N individual
varieties cj,
=
−
NC
PP
c jj
ε
, (5)
where
)]s/1(1/[1
Nj
)s/1(1j
s/1 djcNC−
∈
−−
= ∫ ,
and P is the corresponding price index dual to C; (5) maximises C above subject to the
constraint PCdjcPNj
jj =∫∈
.
Household’s utility is given by4
( ) ( )efP/M,CvV ⋅−= λ , (6)
where, in addition to the usual component v, which we assume to be a Cobb-Douglas
function,
( ) ( )( ) αα
αα
αα −
−
−= 1
1
1P/MCP/M,Cv ,
3 The use of a lump-sum tax is a common simplification in the literature which reduces the distortionary role
of the government. For further explanations see Molana and Moutos (1991), Heijdra and Van der Ploeg (1996) and Heijdra, et al. (1998) among others.
4 For simplicity, like most studies we assume complete separation between households’ and government’s consumption. Therefore, government consumption does not appear in households’ utility function. For some exceptions see, for example, Molana and Moutos (1989), Heijdra et al (1998) and Reinhorn (1998) who extend the original results by allowing for some substitution between the public and private consumption.
13
the utility function also depends on the level of effort, e, that an employed household will
supply when working. ( ) 0ef ≥ captures the disutility of effort; λ=1 for an employed
household; and λ=0 when the household is unemployed. Assuming that ( )ef is taken as
given (see Appendix A.1 for further details on the relevance of these and the derivation of the
effort function) and maximising the utility function of an employed and an unemployed
household subject to their respective budget constraint yields their consumption and money
demand equations. Normalising the number of households to unity and using L and u to
denote the proportions of employed and unemployed households, we have
L+ u=1. (7)
Using this normalisation, the household sector’s aggregate consumption and money demand
equations are
−++−=P
TMuBW)u1(C α , (8)
and
−++−−=P
TMuBW)u1()1(PM α . (9)
Given the above, the aggregate demand for the CES bundle, facing the
monopolistically competitive firms, is Y = C+ G. On the assumption that each firm produces
a distinct variety – given the incentive to specialise due to falling average costs explained
below – the demand function facing firm j is jjj gcy += which is obtained by adding (5)
and (2)5
=
−
NY
PP
ys
jj , (10)
5 We have followed the existing studies in assuming that G and C are similar CES bundles. See Startz (1989),
Dixon and Lawler (1996), Heijdra and Van der Ploeg (1996) and Heijdra, et al. (1998) for further details.
14
since. It is a straightforward exercise to show that P, Y and N satisfy the following
)]s/1(1/[1
Nj
)s/1(1j
s/1 djyNY−
∈
−−
= ∫ , (11)
PYdjyPNj
jj =∫∈
, (12)
and
)s1/(1
Nj
s1j djP
N1P
−
∈
−
= ∫ . (13)
Labour is assumed to be the only factor of production, and to be perfectly mobile
between firms. Firm j’s technology is given by the following production function
φ−= jjj Ley , (14)
where jL is the variable labour input, ej is labour productivity and φ is a constant parameter
reflecting the fixed cost of production assumed to be identical across firms. The increasing
returns to scale implied by falling average cost therefore gives rise to the incentive for full
specialisation from which a one-to-one correspondence between the mass of varieties and
firms in the market results.
We assume that ej is determined by workers’ attitude towards shirking and represents
their optimal effort supply function which depends on: i) the real value of the wage paid by
the firm P/Ww jj = ; ii) the real value of the unemployment benefit, b=B/P, which the
government transfers to the unemployed household; and iii) the extent of unemployment in
15
the economy captured by the unemployment rate6 u. Thus, we postulate the following effort
supply function for a worker employed by firm j
)u,b,w(ee jj = , (15)
which is assumed to satisfy the following properties,
bwas0)u,b,w(e jj ≥≥ ; and bwas0)u,b,w(e jj == .
0we
ej
jjw >=′
∂∂
, 0be
e jjb <=′
∂∂
and 0ue
e jju >=′
∂∂
,
and to have plausible second and cross partial derivatives. In particular, we shall assume
0w
ee
j2
j2
jww <=′′∂∂
, 0wb
ee
j
j2
bwj >=′′∂∂
∂ and 0
wue
ej
j2
uwj >=′′∂∂
∂.
An effort function, which satisfies the above properties, is explicitly derived in Appendix A1.
Each individual firm takes P, Y, N, u and B as given and chooses its ‘efficiency wage’
jW and its price jP so as to maximise its profit
jjjjj LWyP −=π , (16)
subject to the demand function in (10) and the production function in (14) as well as taking
account of its workers’ reaction to the choice of Wj which is given by the effort function in
(15). The first order conditions are 0W/ jj =∂π∂ and 0P/ jj =∂π∂ whose solution imply
the following wage and price setting rules7
jw
jj e
eW
′= , (17)
and
6 Given that the number of households is normalised to 1, u is simply the proportion of unemployed
households and is equivalent to the unemployment rate. 7 The second order conditions are satisfied as long as s>1 and 0e jww <′′ .
16
j
jj e
W1s
sP
−= . (18)
Equation (17) is a well-known result in the efficiency wage literature and implies that
firms raise their wage rate up to the point where the effort function is unit elastic. Equation
(18) is the usual mark-up pricing rule for a monopolistically competitive firm. In a
symmetric equilibrium where all firms are identical, we drop the subscript j and write the
above equations as
ew)u,b,w(ew =⋅′ , (17´)
and
w)u,b,w(e σ= , (18´)
where σ = s/(s-1). To see the (partial equilibrium) implications of these, first note that
together they yield
σ=′ )u,b,w(ew >1. (19)
Totally differentiating (18´) and taking account of (19) implies
dudb
ee
b
u
= −′′
>0, (20)
which shows that an increase in the benefit rate raises the unemployment rate. Totally
differentiating (17´) and (18´) and solving using (20) to eliminate db and dw yields (see
Appendix A2)
′′−′′′′
⋅
′′
= uwb
bwu
ww
eeee
edude σ , (21)
Thus, under our assumptions regarding the shape of the effort function, (21) implies de/du >0
which is consistent with the theoretical consensus that the net result of an increase in
unemployment rate is to raise workers’ effort level.
17
We can use the above results to examine the way in which equilibrium output and
unemployment are related to each other on the supply side. The symmetric equilibrium of the
industry is obtained when entry eliminates profits,
WLPYdjNj
j −== ∫∈
πΠ , (22)
where ∫∈
=Nj
jdjLL is total employment. Thus, through free entry and exit process N adjusts to
ensure Π=0. Imposing this on (22) and solving for Y gives Y = wL, from which upon
substitution for w from (18´) we obtain
eL1Yσ
= . (23)
Equation (23) may be interpreted as a �quasi-aggregate� production function. It traces
the combinations of aggregate employment and output (L,Y) which satisfy the supply side
equilibrium in which labour productivity is determined by an effort supply function and firms
pay wages to induces workers to supply the effort level that maximises their profits. Or, put
differently, these combinations give the equilibrium locus that describes how Y changes as
firms and workers respond to changes in u while the wage is adjusted to ensure the resulting
effort supply maximises profits. Totally differentiating (23) and noting that dL= � du from
(7), we obtain
−
−
= 1
dude
eu
uu1e
dudY
σ. (24)
Thus, provided that de/du in (21) is finite as 1u → , we would expect the right-hand-side of
(24) to be negative for sufficiently large levels of u. Conversely, starting from very low
levels of u, we would expect the right-hand-side of (24) to be positive as long as de/du in (21)
18
is positive, as explained above. Given these and assuming that de/du in (21) is continuous in
u, the equilibrium locus in (u, Y) space may be depicted as in Figure 1 below.
Figure 1. The relationship between output and unemployment with efficiency wages and monopolistic competition
Y Y 1Y 0 lu u hu u→1
The shape depicted in Figure 1 is consistent with our empirical evidence reported in
the previous section which showed that, in seven out of ten countries, dY/du is likely to
change from positive to negative as the level of unemployment passes a certain threshold.
Within the context of the model developed above, the maximum level of output YY = is
achieved at this point where uu = . Prior to this point, effort – and hence productivity – is
relatively lower but is rising with u. Also, the positive effect of a rise in productivity
dominates the negative effect of loss of labour as firms shed workers. Beyond u the opposite
holds and shedding labour that is working efficiently results in a reduction in the level of
output.
It is clear that in the economy described above firms will (rationally) produce the
same level of output, Y1 say, employing either (1-ul) inefficient workers or (1-uh) efficient
workers. Thus, multiple equilibria can arise given the nonlinear nature of the relationship
19
between output and employment8. As a result, the effect of a policy shock on employment
and output depends on the initial equilibrium and unemployment can fall in the event of a
positive shock only if the economy is operating in the efficient region where uu ≥ . Finally, it
can be easily shown that the (tax financed) fiscal multiplier is given by the following
expression (see Appendix A3 for details) where 'DY and '
SY are the slopes of the aggregate
demand and aggregate supply functions in (P, Y) space, respectively,
'S
'D
YY1
1dGdY
−= . (25)
While 'DY < 0 always holds, '
SY can be positive or negative within the framework developed
above. Therefore, the effect of a fiscal expansion depends on the size and sign of the ratio of
the slopes of the two functions. In particular, in the efficient situation when 'SY >0 the
multiplier will – as in most recent new Keynesian studies of the effect of fiscal policy – lie
between zero and unity. But when the economy is operating inefficiently and 'SY <0, the
multiplier can in fact exceed unity – as in the typical Keynesian case – or even become
negative – like in models when more than full crowding out occurs. However, the former
case, in which the rise in output will be accompanied by a fall in employment, corresponds to
an unstable initial equilibrium where – 'SY > – '
DY >0 whereas in the latter case – 'DY >– '
SY >0
and the initial equilibrium is stable.
4. Summary and Conclusions
The main motivating factor underlying our study has been to explore the
circumstances in which positive policy shocks might give rise to adverse employment effects.
8 It is a straightforward exercise to show that the aggregate supply function in (P,Y) space is nonlinear and can
have more than one intersection with the aggregate demand. In such a situation, the equilibrium occurring
20
Accordingly our objective has been two fold. First, we have sought to describe the stylised
facts regarding the co-movement of output and unemployment changes using data from ten
West European countries. Our simple data analysis shows that while in all countries
unemployment changes and output changes can be either positively or negatively related, in
seven of these countries we find a clear (and statistically significant) tendency for
unemployment to be lower when output and unemployment are positively related and higher
otherwise. This suggests a humped-shape relationship between output and unemployment.
Our second objective has been to outline a theoretical setting, which could give rise to such a
relationship. We have therefore constructed a stylised macro-model with goods and labour
market imperfections and have shown that the equilibrium relationship between output and
unemployment can in fact be positive when a rise in unemployment induces workers to
supply a higher effort such that the positive effect of the gain in productivity outweighs the
negative effect of the reduction in employment; otherwise we obtain the conventional result
that output and unemployment are negatively related.
Clearly, our results, which complement those of the literature on the effects of
contractionary fiscal policy (see, for details, Barry and Devereux, 1995), suggest that
plausible circumstances do exist in which market imperfections pose serious obstacles to the
smooth working of expansionary and/or stabilization policies whose final aim is to improve
welfare. We show that the economy can rationally operate at an inefficient equilibrium, and
that positive demand shocks in such circumstances will have perverse effects. Accordingly,
we conclude by stressing Lindbeck’s (1992) concerns about the effectiveness of
macroeconomic stabilisation policy in the presence of labour and product market
imperfection, which are echoed by our empirical and theoretical results.
where the aggregate demand curve is flatter than the (downward sloping) aggregate supply curve will be unstable. See Appendix A3 for further details on the slopes of aggregate demand and supply curves.
21
5. References
Agénor, P-R and J. Aizenman (1999), Macroeconomic adjustment with segmented labor markets, Journal of Development Economics, 58, 277-296.
Aghion, P. and P. Howitt (1992), A model of growth through creative destruction,
Econometrica, 60, 323-351. Aghion, P. and Howitt, P. (1994), Growth and unemployment, Review of Economic Studies,
61, 477-94. Albrecht, J.W and S.B. Vroman (1996), A note on the long-run properties of the shirking
model, Labour Economics, 3, 189-195. Andersen, T.M. and B.S. Rasmussen (1999), Efforts, taxation and unemployment, Economics
Letters, 62, 97-103. Barnes, M. and J. Haskel (2000), Productivity in the 1990s: evidence from British plants,
mimeo. Barry, F. and M. Devereux (1995), The ‘expansionary fiscal contraction’ hypothesis: a neo-
Keynesian analysis, Oxford Economic Papers, 47, 249-264. Blanchard, O.J. and N. Kiyotaki (1987), Monopolistic competition and the effects of
aggregate demand, American Economic Review, 77, 647-66. Blanchard, O.J. (1998), Revisiting European unemployment: unemployment, capital
accumulation and factor prices, mimeo, (MIT). Caballero, R. and M.L. Hammour (1998a), Jobless growth: appropriability, factor
substitution, and unemployment. Carnegie-Rochester Series on Public Policy, 48, 51-94.
Caballero, R. and M.L. Hammour (1998b), The macroeconomics of specificity, Journal of
Political Economy, 103, 724-767. Carter, T.J. (1999), The effect of taxes on labour in efficiency wage models: a comment,
Journal of Public Economics, 72, 325-327. Disney, R., Haskel, J. and Y. Helden (2000), Restructuring and productivity growth in UK
manufacturing, mimeo. Dixon, H. and P. Lawler (1996), Imperfect competition and the fiscal multiplier,
Scandinavian Journal of Economics, 98, 219-231. Fehr, E. (1991), Wages and labor demand: a note, Journal of Institutional and Theoretical
Economics, 147, 539-546. Heijdra, B.J. and F. Van der Ploeg (1996), Keynesian multipliers and the cost of public funds
under monopolistic competition, Economic Journal, 106, 1284-96.
22
Heijdra, B.J., Ligthart J.E., and F. Van der Ploeg (1998), Fiscal policy, distortionary taxation, and direct crowding out under monopolistic competition, Oxford Economic Papers, 50, 79-88.
Leamer, E.E. (1999), Efforts, wages, and the international division of labour, Journal of
Political Economy, 107, 1127-1162. Leith, C. and , C.W. Li (2001), Unemployment and the Productivity Slowdown: A Labour
Supply Perspective, University of Glasgow Discussion Paper in Economics 2001-13. Lindbeck, A. (1992), Macroeconomic theory and the labour market, European Economic
Review, 36, 209-235. Marini G. and P. Scaramozzino (2000), Endogenous growth and social security, CeFiMS DP
02/00/02, Centre for Financial and Management Studies, SOAS, University of London
Malley, J. and T. Moutos (2001), Capital accumulation and unemployment: a tale of two
continents, Scandinavian Journal of Economics, 103, 79-99. Molana, H. and T. Moutos (1989), Useful government expenditure in a simple model of
imperfect competition in Reider, U., Gessner, P. Peyerimoff, A. and Radermacher, F.J. (eds), Methods of Operations Research, 63, Verlag Anton Hain, Frankfurt.
Molana, H. and T. Moutos (1992), A note on taxation, imperfect competition, and the
balanced-budget multiplier, Oxford Economic Papers, 44, 737-52. Molana, H. and J. Zhang (2001), Market Structure and Fiscal Policy Effectiveness,
Scandinavian Journal of Economics, 103, 147-164. Moutos, T. (1991), Turnover costs, unemployment and macroeconomic policies, European
Journal of Political Economy, 7: 1-16. Pisauro, G. (1991), The effect of taxes on labour in efficiency wage models, Journal of Public
Economics, 46, 329-345. Rebitzer, J.B. and L.J. Taylor (1995), The consequences of minimum wage laws; some new
theoretical ideas, Journal of Public Economics, 56, 245-255. Reinhorn, L.J. (1998), Imperfect competition, the Keynesian cross, and optimal fiscal policy,
Economics Letters, 58, 331-337. Sala-i-Martin, X.X. (1996), A positive theory of social security, Journal of Economic
Growth, 1, 277-304. Shapiro, C. and J.E. Stiglitz (1984), Equilibrium unemployment as a worker discipline
device, American Economic Review, 74: 433-44. Smidt-Sørensen, J.B. (1990), The equilibrium effort-wage elasticity in efficiency-wage
models, Economics Letters, 32, 365-369.
23
Smidt-Sørensen, J.B. (1991), An efficiency-wage-hours model and shorter working hours,
Scottish Journal of Political Economy, 38, 113-131. Startz, R. (1989), Monopolistic competition as a foundation for Keynesian macroeconomic
models, Quarterly Journal of Economics, 104, 737-52. van Ark, B., Kuipers, S., and G. Kuper (2000), Productivity, Technology and Economic
Growth, Kluwer Academic Publishers. Yellen, J.L. (1984), Efficiency wage models of unemployment, American Economic Review, 74, 200-205.
24
6. Appendix
A1. Derivation of the Effort Supply Function e(w, u, b)
This appendix explains how a specific effort supply function such as that in equation
(15) can be derived within the framework of the efficiency wage hypothesis where, following
common practice, the agent is assumed to maximise the expected utility of remaining in
employment.
We assume that all households participate in the labour market and at any point in
time a household can be in one of the following states: (i) employed (working); (ii) being
fired (when caught shirking at work); (iii) unemployed (being without a job); or (iv) being
hired (finding a job). Let the utility indices corresponding to of the above states be denoted
as follows:
(i) employed (working): VE
(ii) being fired (losing one’s job): VF
(iii) unemployed (being without a job): VU
(iv) being hired (finding a job): VH
VU and VE can be obtained as follows. Disregarding the money holdings and taxes
(which are the same for all states) and focusing on unemployment benefit or wage as the only
source of work-related income, the utility function in equation (6) implies that the indirect
utility of an unemployed and an employed household is, respectively,
P/BV U = , (A1.1)
and
( ) )e(fP/WV E −= . (A1.2)
While (A1.1) is straightforward, (A1.2) needs some explanation regarding f(e). We shall
assume 0f >′ and 0f ≥′′ which implies that the disutility of effort rises with a non-
25
decreasing rate. In particular, we shall use the explicit form 2ke)e(f = where k>0 is a
scaling factor.
Finally, we need to specify VH, which is the satisfaction a household attaches to
finding a job or being hired. But the utility associated with this state is in principle not
distinguishable from VE and for simplicity we let
VH = VE, (A1.3)
The probabilities associated with moving from one state to another are assumed to be
determined as follows:
(a) Probability associated with being fired when shirking, F.
We assume that shirking is the only reason for being fired (we do not explicitly model the
monitoring technology). Therefore, ceteris paribus, F is a monotonic function of the effort
level, e. Thus,
( ) ( ) ( ) 0dedF;01F;10F;eFF <=== .
For simplicity, normalise the maximum possible effort to unity and let
F = 1 � e. (A1.4)
(b) Probability associated with finding a job, or being hired, when unemployed, H.
We assume that the labour force is homogeneous and, ceteris paribus, H is a monotonic
function of the unemployment rate, u (we do not explicitly model the search technology).
Thus,
( ) ( ) ( ) 0dudH;01H;10H;uHH <=≤= .
For simplicity we let
H = 1 � u. (A1.5)
26
We define the optimal level of effort as that which maximises a household’s expected
utility of remaining in employment. The latter is denoted by R(e) and is, by definition, given
by
R(e) = (1-F)VE + FVF. (A1.6)
Also, given that a ‘fired’ worker can either be hired or remain unemployed, we let VF be a
weighted average of VH and VU. Thus,
VF = HVH + (1-H)VU. (A1.7)
Equations (A1)-(A7) yield
( ) ( )( )ubkew)u1()e1(kewe)e(R 22 +−−−+−= ,
where w=W/P and b=B/P. This equation can be rearranged as
( )ubw)u1(e)bw(uke)u1(uke)e(R 23 +−+−+−−−= . (A1.8)
The agent takes (w, b, u) as given and chooses e to maximise R(e). The first order
condition for this is ( ) ( ) 0)bw(k3/1eu3/)u1(2uke3 2 =−+−−− . This has two roots of
which only one is positive which also satisfies the second order for a maximum and can, after
some normalisation, be written as
uu1
uu1)bw(e
21
−−
−+−= γ , (A1.9)
where γ ≡3/k. It is clear that equation (A1.9) satisfies our specified conditions, since
bwas0)u,b,w(e ≥≥ ; bwas)1,0(u0)u,b,w(e =∈∀= ; 0weew >=′
∂∂ ;
0beeb <=′
∂∂ ; 0
ueeu >=′
∂∂ ; 0
wee 2
2
ww <=′′∂∂ ; 0
wbee
2
bw >=′′∂∂
∂ ; and 0wuee
2
uw >=′′∂∂
∂ .
27
A2. Derivation of Equations (20) and (21).
We use the following equations (15), (17´), (18´) and (19) which are reproduced
below as (A2.1)-(A2.4), respectively,
)u,b,w(ee = , (A2.1)
ew)u,b,w(ew =⋅′ , (A2.2)
w)u,b,w(e σ= , (A2.3)
σ=′ )u,b,w(ew . (A2.4)
First, totally differentiating (A2.3) yields
dwduedbedwe 'u
'b
'w σ=++ , (A2.5)
and substituting from (A2.4), i.e. σ=′we , in (A2.5) we obtain
0duedbe 'u
'b =+ , (A2.6)
which is solved to yield equation (20).
Next, totally differentiating (A2.4) implies
0duedbedwe "uw
"bw
"ww =++ , (A2.7)
and using (A2.6) to eliminate db we have
0duedueeedwe "
uw'b
'u"
bw"ww =+
−+ , (A2.8)
which can be solved for dw to give
dueeee
e1dw "
uw'b
'u"
bw"ww
−
= , (A2.9)
Substituting from (A2.9) into de = σ dw implied by (A2.3), we obtain equation (21).
28
A3. Derivation of the Fiscal Multiplier, Equation (25).
We derive the fiscal multiplier as follows. First, the aggregate demand function (AD)
is derived by noting that Y = C + G, where C is obtained by solving equations (8) and (9),
i.e. [ ]( )P/M)1/(C αα −= . Hence,
PM
1GY
−+=
αα (A3.1)
Totally differentiating (A3.1), for any give M , then implies that on AD,
dPP)1(
MdGdY 2
−
−=α
α . (A3.2)
Next, recalling from equations (23) and (7) that Y = wL and L= 1-u, we obtain, on the
aggregate supply (AS) side, Y = w(1-u), which can be totally differentiated to yield
dY = (1-u)dw � wdu, (A3.3)
which upon substitution from (A2.9) can be written as,
duweeee
eu1dY "
uw'b
'u"
bw"ww
−
−
−= . (A3.4)
Also, using (A2.6) and the fact that for any given B, dPPBdb 2−= , we obtain
dPPB
eedu 2'
u
'b
= , (A3.5)
which can be substituted in (A3.4) to give the reaction of output to a change in the price level
on the aggregate supply (AS),
dPPB
eewe
eee
eu1dY 2'
u
'b"
uw'b
'u"
bw"ww
−
−
−= . (A3.6)
Simplifying notation and writing (A3.2) and (A3.6) as
29
dPYdS
dGdPYdY
'S
'D
=
+=
where 'DY and '
SY are the slopes of AD and AS in (P, Y) space. Solving the above to
eliminate dP we obtain the fiscal multiplier in equation (25), namely,
'S
'D
YY1
1dGdY
−= . (A3.5)